You are not logged in.
My older Arch-install (0.7 something) had speeds of around 50Mbps copying large files TO the server.
Now, with a fresh install of newest Arch, with exactly the same smb.conf, on exactly the same hardware, I get 30-40 Mbps.
Only real change I did this time was choosing jfs filesystem, since people told me it was really safe when the power goes down, not messing filesystem up.
Could it be this filesystem, with its increased security, is slower someway?
Filecopy FROM server to windows-machine is as fast as before...60-70Mbps or so.
Maybe its normal speeds anyway, XP -> Linux, I dont feel I have time to bother with it. Just thought it was a bit strange, since I have the exact same smb.conf.
Offline
This won't help you with your samba issue, but not one single filesystem is safe when the power goes down. You best bet might be ext3+data=journal, mounted 'sync'. But forget about speed then.
1000
Offline
Well, I've never had anything ruined (that I noticed) in windows, but with Linux it often happens stuff.
Anyway, some people told me that jfs should be real safe, and that they never had anything ruined when powered down abruptly.
So, why not I thought, and tested it this time I installed. Guess it wont do any harm compared to ext 3 I had before....except maybe its a bit slower then?
Offline
This topic on tuning the ext3 file system may be quite helpful. Full journaling is not really slow after all.
---for there is nothing either good or bad, but only thinking makes it so....
Hamlet, W Shakespeare
Offline
But first of all, is it realistic that jfs itself is so slow that copying TO the server could be as slow as 30-40Mbps?
I mean, that is really really slow, if it is the filesystem that is limiting. Must be some extreme security to jfs if that alone is responsible for this downfall in throughput in my network at home...
Offline
all this is file with ext3, except that it seems that ext3 has/had some special issues with samba. I don't remember topic title but some recently run and published on lkml benchmarks suggest that ext3/4 make samba slower than other linux fs. This did not seem to be a difficult to fix issue, but I don't know if this problem was fixed already or not.
Offline